Merge pull request #354 from mgeisler/indent-benchmark

Make `indent` faster by preallocating the output string

Cargo.toml

@@ -20,6 +20,11 @@ name = "linear"
 harness = false
 path = "benches/linear.rs"
 
+[[bench]]
+name = "indent"
+harness = false
+path = "benches/indent.rs"
+
 [features]
 default = ["unicode-linebreak", "unicode-width", "smawk"]
 

benches/indent.rs

@@ -0,0 +1,24 @@
+use criterion::Criterion;
+use criterion::{criterion_group, criterion_main};
+
+pub fn benchmark(c: &mut Criterion) {
+    // Generate a piece of text with some empty lines.
+    let words_per_line = [
+        5, 10, 15, 5, 0, 10, 5, 0, 5, 10, // 10 lines
+        10, 10, 5, 5, 5, 5, 15, 10, 5, 0, // 20 lines
+        10, 5, 0, 0, 15, 10, 10, 5, 5, 5, // 30 lines
+        15, 5, 0, 10, 5, 0, 0, 15, 5, 10, // 40 lines
+        5, 15, 0, 5, 15, 0, 10, 10, 5, 5, // 50 lines
+    ];
+    let mut text = String::new();
+    for (line_no, word_count) in words_per_line.iter().enumerate() {
+        text.push_str(&lipsum::lipsum_words_from_seed(*word_count, line_no as u64));
+        text.push('\n');
+    }
+    assert_eq!(text.len(), 2304); // The size for reference.
+
+    c.bench_function("indent", |b| b.iter(|| textwrap::indent(&text, "    ")));
+}
+
+criterion_group!(benches, benchmark);
+criterion_main!(benches);

src/indentation.rs

@@ -50,7 +50,11 @@
 /// assert_eq!(indent(" \t  Foo   ", "->"), "-> \t  Foo   ");
 /// ```
 pub fn indent(s: &str, prefix: &str) -> String {
-    let mut result = String::new();
+    // We know we'll need more than s.len() bytes for the output, but
+    // without counting '\n' characters (which is somewhat slow), we
+    // don't know exactly how much. However, we can preemptively do
+    // the first doubling of the output size.
+    let mut result = String::with_capacity(2 * s.len());
     let trimmed_prefix = prefix.trim_end();
     for (idx, line) in s.split_terminator('\n').enumerate() {
         if idx > 0 {